This pdf of a PPT presentation seems to address your question well. A yearly probability is a rate in its parlance. Your model needs to use the probaility in a time period, which can be computed from that rate.

https://www.hsrd.research.va.gov/for_researchers/cyber_seminars/archives/819-notes.pdf

For a more academic citation, this article may provide the procedure you need.  

Diaby V, Adunlin G, Montero AJ. Survival modeling for the estimation of transition probabilities in model-based economic evaluations in the absence of individual patient data: a tutorial.  Pharmacoeconomics. 2014 Feb;32(2):101-8. doi: 10.1007/s40273-013-0123-9.

.

Dear Dr Miller,

Thank you for your reply. The presentation is indeed very useful and I will go through the paper as well. I think I have the survival probabilities obtained from Literature and not the survival rate. Please have a look at the table attached. 

Can these be used as the transition probabilities?

Regards,

Karan

Dear Karan Shah,

I attached one manual of the software TreeAge, this manual shows how to transform odds to probabilities. Also they show another types of calculations that you may use.

You can use this data for the transition probabilities, but you will need to adjust to your model. 

I propose using partitioned survival models

good luck

https://www.ispor.org/ScientificPresentationsDatabase/Presentation/59159

Hi Karan,

If the cycle length in your model will be one year then annual survival probabilities you already have can be used to inform the transition from alive to death in your model. If for some reason you want to change cycle length then appropriate mathematical transformations will be needed. The only caution is needed here: "I have the yearly survival probabilities which were calculated based on the odds ratios of the medications" - is this transformation is correct. 

p.s.

And Ahmed is suggesting the good approach with partitioned survival models, which can be better solution then Markov in this case.